Generally, digital conference arrangements combine information samples such as, for example, a speech sample from each conference port during some time period, and then distribute the resulting sample to each conferee. In a teleconference the communication between conferees should sound like a typical two-party connection; however, this is usually not the case when the participants communicate with each other over various types of transmission facilities which introduce irregularities such as, for example, noise, echo, and other types of distortion. For example, participants commonly access a conference via a telephone line which forms one two-way communication path. A digital conference circuit, however, requires two one-way communication paths. The transition between these two types of paths is usually made by a hybrid circuit which causes signal reflections, perceived as echo, on the lines. When the conferees are closely located, this echo is usually not objectionable to the listener. However, automatic gain control circuits, which adjust the speech level of a participant, amplify the echo and may cause the arrangement to emit a high-pitched tone particularly when there are multiple feedback paths due to several conferees being allowed to speak simultaneously. This high-pitched tone called singing, is quite annoying to the participants and prevents further conversation on the arrangement.
One of the more common ways to minimize echo and singing is to allow only one conferee to speak at any given time. This can be accomplished on a first-come, first-serve basis or by using the highest level of speech to control which conferee is heard. Both arrangements have disadvantages in that each allows one speaker to dominate the conference. Also, in the latter case, sudden loud bursts of conversation from others or noise, such as the rustling of papers can break up the current speaker's conversation, thereby losing intelligibility. Furthermore, since the speech level of each participant is a function of distance and equipment, some participants may never get to speak on the arrangement at all.
In contrast to the one speaker arrangements, there are those systems which allow all of the conferees to speak simultaneously. These conference circuits usually involve inserting loss in the receive path of each conferee or speaker. As a result, the amount of feedback as well as the speech level to each conferee is reduced. If automatic gain control is employed to boost the speech level of each speaker, the probability of singing increases, and as a result, additional loss must be inserted. Other arrangements which permit simultaneous speakers disconnect the transmit path of each conferee until the conferee's speech level reaches a threshold. While this approach eliminates a feedback path when a conferee is silent it still may cause singing if too many conferees speak simultaneously.